Reduction of RA Signaling Causes an Increase in the Number of Atrial and Ventricular Cells

(A–C) Frontal views of hearts at 48 hpf expressing the transgene Tg(cmlc2:DsRed2-nuc) (red). Atria are labeled with the anti-Amhc antibody S46 (green). (B and C) Treatment with 1 μM DEAB or 10 μM BMS causes cardiac chamber enlargement.

(D) Mean and standard deviation (SD) of numbers of atrial and ventricular cardiomyocytes. Stages indicate when DEAB or BMS treatment was initiated. Asterisks indicate statistically significant differences relative to wild-type (p < 0.005).

(E) Fold difference in mean cell numbers relative to wild-type (WT) controls.

(F–I) Reduction of RA signaling causes increased expression of amhc and vmhc. Dorsal views, anterior to the top at the 20 or 22 somite stage.

(J) Mean and SD of areas of expression of amhc and vmhc in WT and BMS-treated embryos. Asterisks indicate statistically significant differences relative to WT (p < 0.005).

(K) Fold difference in mean areas of gene expression relative to WT controls.

(L and M) Schematic of 40% epiboly fate maps from WT and BMS-treated embryos indicating that the relative position of regions containing APs (green) and VPs (red) is unchanged when RA signaling is reduced. Percentages indicate the frequency of encountering APs or VPs in each region.

(N) Reduction of RA signaling increased the frequency of encountering VPs, but not APs, at 40% epiboly. The asterisk indicates statistically significant difference relative to WT (p < 0.005).

(O) Reduction of RA signaling increased the mean number of labeled progeny produced by APs but not VPs. The asterisk indicates statistically significant difference relative to WT (p < 0.05).

RA Signaling Positively Regulates hoxb5b Expression in the LPM

Expression of hoxb5b at the tailbud (tb) or 8 somite stages, dorsal views, anterior to the top.

(A–D and F–I) Reduction of RA signaling inhibits hoxb5b expression, and treatment with RA induces ectopic hoxb5b expression.

(E and J) Expression of the hoxb5b paralog hoxb5a. Identification of hoxb5b as a RA-responsive gene is not surprising, since the chick and mouse Hoxb5 orthologs and the zebrafish paralog hoxb5a have already been implicated as direct targets of RA signaling ([Bruce et al., 2001], [Grandel et al., 2002], [Oosterveen et al., 2003] and [Sharpe et al., 1998]).

RA-Responsive Genes Are Expressed in the Forelimb Field

(A–E) Expression of nkx2.5, gata4, retd, hoxb5b, and raldh2 relative to ntl, which marks the notochord; dorsal views of flatmounted embryos, anterior to the top, at the 8 somite stage. Brackets and numbers indicate the mean distance of the limit of gene expression relative to the tip of the notochord, measured in notochord cell diameters (ncds). (A and B) The posterior portions of nkx2.5 and gata4 expression overlap with the notochord. (C) retd expression is found posterior to nkx2.5 and gata4 expression. (D and E) hoxb5b is expressed closer to raldh2 expression in the somites, a major source of RA in the embryo.

(F) Schematic of the WT LPM fate map at the 8 somite stage, indicating locations of atrial (green, A), ventricular (red, V), and fin (blue, F) progenitors relative to the anterior-posterior axis of the notochord, measured in ncds. Numbers in the table represent the number of labeling experiments in which each progenitor type was detected at a certain ncd position; n indicates the total number of experiments at each position. Experiments were performed on both the left and right sides of the embryo; no asymmetries were observed.

(G–I) When RA signaling is reduced, FPs are absent and CPs frequently occupy FP territory.

(J) In hoxb5b morphants, FPs are present and CPs rarely occupy FP territory.

Altered Expression of gata4 and tbx5 in Embryos Deficient in RA Signaling

(A–D) Expression at the 8 somite stage of gata4 and ntl in WT, BMS-treated, DEAB-treated, and RO-treated embryos; dorsal views, anterior to the top. When RA signaling is reduced, the total extent of gata4 expression is longer (blue arrows). This shift in gata4 expression is most prominent posterior to the notochord tip (black arrow).

(E–H) Expression of tbx5 in WT, BMS-treated, DEAB-treated, and RO-treated embryos at the 12 somite stage; lateral views, anterior to the left. (E) In WT embryos at this stage, the total extent of tbx5 expression (yellow outline) is divided into two comparably sized populations; a black arrow indicates the separation point. (F–H) When RA signaling is reduced, the separation between populations is less evident.

Mosaic Analysis Demonstrates a Cell-Autonomous Requirement for RA Signaling during Forelimb Formation

(A–F) Experimental design for mosaic analysis. (A) Donor embryos were injected with fluorescein dextran at the one-cell stage. (B and C) At sphere stage, blastomeres from WT, Tg(hsp70:dnRARα), or BMS-treated donor embryos were transplanted into the margin of a WT host embryo. (D) At 48 hpf, progeny of donor-derived cells (arrows) were detected in the heart and fin by anti-fluorescein immunohistochemistry. (E) Mosaic embryo containing donor-derived cardiomyocytes. (F) Mosaic embryo containing donor-derived cells in pectoral fin. Only LPM-derived fin cartilage cells (yellow arrows), and not fin ectoderm (black arrow) or somite-derived fin muscle ([Haines and Currie, 2001] and [Neyt et al., 2000]), were scored as fin.

(G and H) Frequency of heart and fin contributions from heat-shocked Tg(hsp70:dnRARα), non-heat-shocked Tg(hsp70:dnRARα), WT, or BMS-treated donor-derived cells. Percentages reflect the number of mosaic embryos exhibiting donor-derived fin mesenchyme (F), atrial (A), or ventricular (V) cardiomyocytes relative to the total number of transplantation (TT) experiments. Asterisks indicate statistically significant differences from WT (p < 0.05).

hoxb5b Morphants Display an Increased Number of Atrial Cardiomyocytes

(A) MOs targeting the donor site of the single intron in either hoxb5a or hoxb5b efficiently abrogate splicing of the respective transcript without affecting the other transcript. Larger bands in lanes 2 and 4 represent the retained introns of hoxb5a and hoxb5b, respectively.

(B–E, G, and H) hoxb5a morphants have no evident morphological defects, and hoxb5b morphants exhibit enlarged hearts, pericardial edema, and overtly normal pectoral fins. Lateral views, anterior to the left, 48 hpf, with higher magnification of pectoral fin outlined in (E), (G), and (H). (F) Genomic intron-exon structure is completely conserved and nucleotide sequences are highly conserved in the locations that we targeted with anti-hoxb5b and anti-hoxb5a MOs. Uppercase letters designate first exon sequence, and lowercase letters are intronic sequence. Underline indicates the respective MO target sequences. Asterisks indicate sequence differences.

(I–R) hoxb5b morphants display an increased number of atrial cardiomyocytes and a normal number of ventricular cardiomyocytes. Views and graphs are as presented in Figure 1. Error bars represent SD.

Hoxb5b Acts Nonautonomously to Restrict Atrial Cell Number

(A–D) Experimental design for transplantation of WT donor cells into morphant host embryos. (A) WT donor embryos carrying Tg(cmlc2:DsRed2-nuc) were injected with fluorescein dextran at the one-cell stage. (B and C) At sphere stage, donor blastomeres were transplanted into the margin of hoxb5a or hoxb5b morphant host embryos. (D) At 48 hpf, cardiomyocyte progeny of donor-derived cells (arrows) were detected on the basis of their nuclear DsRed localization. Fixation destroys the fluorescence of the fluorescein dextran lineage tracer, allowing for clean visualization of Amhc (green).

(E) Example of donor cells (arrows) contributing to both the atrium and the ventricle in a hoxb5a morphant host.

(F) Example of donor cells (arrows) contributing to the atrium in a hoxb5b morphant host.

(G) Frequency of donor cell contributions to each cardiac chamber in hoxb5a and hoxb5b morphant hosts.

(H–K) Experimental design for transplantation of hoxb5b morphant or mRNA-injected donor cells into morphant host embryos. (H) WT donor embryos were injected with hoxb5b MO or hyperactive hoxb5b mRNA along with fluorescein dextran at the one-cell stage. (I and J) At sphere stage, donor blastomeres were transplanted into the margin of hoxb5b morphant host embryos carrying the transgene Tg(cmlc2:DsRed2-nuc). (K) At 48 hpf, we selected host embryos in which donor-derived cells contributed to anterior mesoderm lineages other than the heart and counted the number of cardiomyocytes in each of these hosts.

(L and M) Donor-derived cells expressing hyperactive hoxb5b nonautonomously reduce the number of atrial cells in hoxb5b morphant embryos. Asterisks represent statistically significant differences between the pair of values indicated by each bar. Error bars represent SD.

(N–P) Model of interactions between the forelimb field and heart field that restrict the number of CPs. (N) RA signaling acts on the forelimb field to promote formation of FPs (blue) and expression of RA-responsive genes, including hoxb5b (yellow oval). This indirectly results in production of two hypothesized repressive signals that limit formation of (1) APs (green) and (2) VPs (red). (O) In the absence of RA signaling, the expanded AP and VP populations occupy the space created by the loss of FPs. (P) Without hoxb5b, hypothesized to regulate repressive signal 1, FPs are intact and still restrict VP formation, but the AP population is expanded.

The Numbers of atrial and Ventricular Cells Are Increased in raldh2 Morphants and neckless (nls/raldh2) Mutants
All views and graphs are as presented in Figure 1.
(A and B, F and G) raldh2 morphants exhibit increased atrial and ventricular cell number; see also Table S3. Asterisks indicate statistically significant differences relative to wild-type (p<0.01).
(C–E and H–J) nls mutants have increased expression of amhc and vmhc; see also Supplemental Table 4. Asterisks indicate statistically significant differences relative to wild-type (p<0.005). The phenotypes of raldh2 morphants and nls mutants are less dramatic than those observed in BMS-treated embryos (Figure 1), most likely because of limits on MO efficacy and the hypomorphic nature of the nls allele (Begemann et al., 2001).

Treatment with the RARα Antagonist Ro41-5253 (Lopez-Boado et al., 1996; Shibakura et al., 1997; Zhang et al., 1995) Phenocopies Treatment with More General Antagonists of RA Signaling, Suggesting that RARα Paralogs Are Required to Promote Forelimb Formation and Limit Cardiac Cell Number
(A and B, E and F) Expression patterns of rar parlogs at the tailbud stage; dorsal views, anterior to the top. (A and B) rarα paralogs (blue) (Hale et al., 2006; Waxman and Yelon, 2007) are expressed in posterior portions of LPM near raldh2 expression (redbrown) and therefore appear to be the best candidates for mediating RA signaling in FPs.
(E and F) rarγ paralogs (blue) are expressed more anteriorly and in the tailbud, farther from raldh2 expression.
(C and D, G and H) Treatment with RO causes phenotypes similar to those of embryos deficient in RA signaling, including pericardial edema and loss of pectoral fins; lateral views, anterior to the left, at 48 hpf, with higher magnification of pectoral fin outlined in (D and H).
(I and N) tbx5 expression at 30 hpf; dorsal views, anterior to the top. RO-treated embryos lack tbx5 expression in the pectoral fin bud mesenchyme.
(J–M and O–R) RO treatment does not affect retd expression at the tailbud stage, but reduces retd expression at the 8 somite stage; in contrast, RO treatment results in loss of hoxb5b expression at both the tailbud and 8 somite stages. Views are as presented in Figure 2. (S-b) RO treatment causes increases of atrial and ventricular cardiomyocyte populations comparable to those observed in DEAB-treated embryos; see also Tables S1 and S2. Views and graphs are as presented in Figure 1.

Fate Map Comparisons Indicate that RA Signaling Limits the Number of Atrial and Ventricular Cardiomyocytes through Independent Mechanisms
(A) Schematic of an embryo at 40% epiboly, lateral view, dorsal to the right. Blue grid indicates the coordinates of the marginal region of the embryo represented in the fate maps. The vertical axis indicates the cell tier (latitude), the distance in cell diameters from the embryonic margin. The horizontal axis indicates the angular distance (longitude) from the dorsal midline (0°:) in a clockwise or counterclockwise direction along the embryonic circumference.
(B and C) Cardiomyocyte progenitor fate maps from wild-type embryos (Keegan et al., 2004) (B), and BMS-treated embryos (Keegan et al., 2005) (C). The portion of the embryonic margin containing cardiomyocyte progenitors overlaps with the expression of raldh2 in lateral and ventral mesendoderm at 40% epiboly (Begemann et al., 2001; Grandel et al., 2002; Keegan et al., 2004). Each circle represents the location of labeled cells in an individual experimental embryo. Green and red circles indicate that labeled cells gave rise to atrial or ventricular cardiomyocyte progeny, respectively. Brown circles indicate experiments in which labeled cells did not give rise to cardiomyocyte progeny. Green and red backgrounds indicate territories where atrial and ventricular progenitors, respectively, are found in both the wild-type and BMS-treated fate maps (schematized in Figures 1L and 1M). Striped background indicates territories containing atrial and ventricular progenitors. Atrial progenitors arose from similar territories in wild-type and BMS-treated embryos: in wild-type embryos, blastomeres giving rise to atrial progeny were found between 90°:-135°: in tier 2, between 100°:-140°: in tier 3, and between 105°:-110°: in tier 4, while in BMS-treated embryos, they were found between 100°:-135°: in tier 2, between 100°:-160°: in tier 3, and between 115°:-120°: in tier 4 (see also Supplemental Table 3). However, our data suggested that there may be a slightly broader distribution of ventricular progenitors in BMS-treated embryos than in wild-type embryos: in wild-type embryos, blastomeres giving rise to ventricular progeny were found between 60°:-125°: in tier 1, between 60°:-120°: in tier 2, and between 90°:-95°: in tier 3, whereas in BMS-treated embryos, they were found between 50°:-150°: in tier 1, between 60°:-125°: in tier 2, and between 65°:-115°: in tier 3 (see also Supplemental Table 3). Even considering this modest expansion, the territories containing atrial and ventricular progenitors appeared comparable in wild-type and BMS-treated embryos at 40% epiboly. Additionally, reduction of RA signaling does not disrupt the frequency of encountering atrial progenitors within the green territories (see Figure 1N). However, reduction of RA signaling does cause a marked increase in the frequency of encountering ventricular progenitors within red territories (see Figure 1N).
Wild-type fate map data were presented previously in a different format (Keegan et al., 2004). A portion of the BMS189453 fate map data was presented previously in a different format (Keegan et al., 2005), but this report did not include examination of atrial and ventricular lineages.

RA Signaling Positively Regulates retd in the LPM
Expression of retd at the tailbud (tb) or 8 somite stages, dorsal views, anterior to the top. (A-D and F-I) Reduction of RA signaling inhibits retd expression, and treatment with RA induces ectopic retd expression. (E) At tb, retd expression (blue) overlaps with raldh2 expression (red-brown; arrow). (J) By the 8 somite stage, retd expression is anterior to the expression of raldh2 in the somites (arrow). krox20 expression provides a marker of hindbrain rhombomeres 3 and 5 (anterior brown stripes; arrowheads). RA treatment is known to positively regulate retd expression in cell culture (Cerignoli et al., 2002), but it has not been previously shown that RA signaling is required for its expression in the embryo nor has its expression been previously reported in zebrafish. Based on its proximity to raldh2 and three retinoic acid response elements in the 2 kb 5′ to the retd start site (J.S.W. and D.Y., unpublished data; Bastien and Rochette-Egly, 2004), it is feasible that retd is a direct target of RA signaling.

Comparison of the Expression Patterns of nkx2.5, gata4, retd, hoxb5b, and raldh2
Dorsal views, anterior to the top.
(A and B) nkx2.5 (blue in A) and gata4 (blue in B) are expressed at a distance from raldh2 (red-brown) in the somites at the 8 somite stage.
(C and D) At the 8 and 12 somite stages, after the requirement for RA signaling to limit cardiac cell number has primarily passed (Figures 1D and 1E), there are low levels of raldh2 expression in the posterior LPM (arrows) and the otic vesicles (arrowheads). The staining reaction for these particular in situ hybridizations ran significantly longer than all others represented in our figures. Prior to the 8 somite stage, even particularly long staining reactions did not reveal raldh2 expression in the LPM, consistent with prior reports (Begemann et al., 2001; Grandel et al., 2002; JSW and DY, unpublished data).
(D–F) At the 6 somite stage, when robust expression of nkx2.5 and gata4 begins, these genes are found just anterior to the RA-responsive gene retd. Arrows in (E,F) indicate the anterior limit of retd expression.
(G–I) By the 8 somite stage, expression of cardiac marker genes nkx2.5 and gata4 is separated from retd expression. Arrows in (H,I) indicate the space between expression domains.
(J–L) At the 8 somite stage, expression of nkx2.5 and gata4 is at a distance from hoxb5b expression.

Procedure for Fate Mapping at the 8 Somite Stage
(A) 5-10 cells within the LPM were labeled, and their position relative to the tip of the notochord was recorded. Image is an overlay from a representative experiment in which cells were labeled at 70 ncds.
(B) At 48 hpf, progeny of labeled cells are detected.
(C) WT embryo exhibiting labeled cells in pectoral fin. Presumptive fin cartilage (yellow arrow) was scored as fin, while fin ectoderm (black arrow) was not.
(D) BMS-treated embryo exhibiting labeled cardiomyocytes (arrows).

Activation of the Transgene Tg(hsp70:dnRARα) Causes Phenotypes Consistent with a Loss of RA Signaling
(A–D) Heat shock of embryos carrying the transgene Tg(hsp70:dnRARα) causes pericardial edema (arrowhead in C), shortened hindbrain (arrow in C), and loss of pectoral fins (outlined in B); lateral views, anterior to the left, of live embryos at 48 hpf, with higher magnification of pectoral fin region shown in (B,D).
(E and H) tbx5 expression at 30 hpf; dorsal views, anterior to the top. In heat shocked transgenic embryos, tbx5 expression is lost from the forelimb region.
(F and I) cmlc2 expression at 22 hpf; dorsal views, anterior to the top. Heat shocked transgenic embryos display a modest expansion of cmlc2 expression.
(G and J) At 48 hpf, hearts of heat-shocked transgenic embryos are modestly dilated and exhibit an increase in atrial cell number; see also Table S1. Images and cell counts were generated by crossing a fish carrying the transgene Tg(hsp70:dnRARα) with a fish carrying the transgene Tg(cmlc2:DsRed2-nuc) . Both non-heat shocked, transgenic siblings and non-transgenic, heat shocked siblings were morphologically normal and considered wild-type.

Embryos Injected with hoxb5b RNA or vp16-hoxb5b RNA Display Similar Phenotypes
(A–C) Lateral views, anterior to the left, of live embryos at 48 hpf.
(A) WT embryo.
(B) Embryo injected with 150 pg of hoxb5b RNA.
(C) Embryo injected with 30 pg of vp16-hoxb5b RNA. Both hoxb5b RNA and vp16- hoxb5b RNA posteriorize the embryo, resulting in the loss of anterior structures, like the eyes (arrows in B and C). However, vp16-hoxb5b RNA is more potent and can achieve the same phenotypes as hoxb5b RNA at one-fifth the dose.